Activity and stability of different Fe loaded primary catalysts for tar elimination

Fuel 317 (2022) 123457 Available online 8 February 2022 0016-2361/© 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Activity and stability of different Fe loaded primary catalysts for tar elimination Maria Cortazar a , Jon Alvarez b , * , Leire Olazar a , Laura Santamaria a , Gartzen Lopez a , c , Heidi Isabel Villaf´ an-Vidales d , Asier Asueta e , Martin Olazar a a Department of Chemical Engineering, University of the Basque Country UPV/EHU, P.O. Box 644-E48080, Bilbao, Spain b Department of Chemical and Environmental Engineering, University of the Basque Country UPV/EHU, Nieves Cano 12, 01006 Vitoria-Gasteiz, Spain c IKERBASQUE, Basque Foundation for Science, Bilbao, Spain d Universidad Nacional Autonoma de Mexico, Instituto de Energias Renovables, Temixco, Morelos, Mexico e GAIKER Technology Centre, Basque Research and Technology Alliance (BRTA), Parque Tecnol´ ogico de Bizkaia, edificio 202, E-48170 Zamudio, Spain ARTICLE INFO Keywords: Primary catalyst Biomass gasification Tar model compound Tar elimination Fluidized bed Catalyst deactivation ABSTRACT The performance of olivine, dolomite and γ-alumina primary catalysts was evaluated in the continuous tar elimination process in which toluene was selected as the biomass gasification tar model compound. Iron was incorporated into these catalysts in order to improve their catalytic activity. All the experiments were performed in a continuous flow fluidized bed micro-reactor, with a steam/toluene ratio of 4 and a space velocity (GHSV) of 820 h −1 , which corresponds to a catalyst amount of 3.8 cm 3 . The effect of temperature was studied using olivine in the 800–900 ◦C range, which allowed concluding that 850 ◦C was the best temperature for tar removal. The fresh and deactivated catalysts were characterized by N 2 adsorption–desorption, X-ray fluorescence (XRF), X-ray diffraction (XRD) and temperature-programmed oxidation (TPO). Tar conversion efficiency was assessed by means of carbon conversion, H 2 yield (based on the maximum allowed by stoichiometry), gas composition and product yields, with Fe/Al 2 O 3 leading to the highest conversion (87.6 %) and H 2 yield (38 %). Likewise, Fe/ Al 2 O 3 also provided the highest stability, as it allowed operating for long periods with high conversion values (85.9 % after 35 min on stream), although it underwent severe deactivation. The analysis of the spent catalysts revealed that deactivation occurred mainly by coke deposition on the catalyst surface and iron phase oxidation, with Fe/olivine and Fe/dolomite leading to the faster deactivation due to their poorer metal dispersion related to their reduced surface area. The TPO profiles showed that the coke deposited on the three catalysts was amorphous with a very small contribution of highly structured carbon. 1. Introduction Biomass gasification is considered one of the most efficient routes to convert biomass feedstock into gaseous fuel through a partial oxidation process at high temperatures [1,2]. However, one of the main shortcomings of biomass gasification lies in the presence of tars in the product stream, which leads to fouling/clogging and corrosion of downstream equipment [3–5]. Hence, in order to minimize the amount of tar and improve the syngas composition, its catalytic conversion is one of most promising routes [6,7]. This process involves the oxidation of the tar components using steam to produce a syngas richer in H 2 and, furthermore, the presence of the catalyst allows a more effective tar removal at lower temperatures than those in the non-catalytic tar conversion [8]. The tar is as a complex mixture of condensable hydrocarbons, ranging from single-ring to five-ring aromatic compounds along with other oxygen-containing hydrocarbons and complex polycyclic aromatic hydrocarbons (PAH) [9]. Tar model compounds have been widely used in order to ascertain the catalyst performance and determine suitable operating conditions. In this work, toluene was selected as a tar model compound because it is a stable aromatic structure, especially at relatively low temperatures, apart from being one of the major tar species in the biomass gasification [10–14]. Many parallel and consecutive reactions involve tar conversion, with the product distribution being the result of their competition. The main products obtained are hydrogen, carbon monoxide and carbon dioxide, and the major reactions occurring in the process are as follows: * Corresponding author. E-mail address: [email protected] (J. Alvarez). Contents lists available at ScienceDirect Fuel journal homepage: www.elsevier.com/locate/fuel https://doi.org/10.1016/j.fuel.2022.123457 Received 23 November 2021; Received in revised form 21 January 2022; Accepted 26 January 2022 Fuel 317 (2022) 123457 2 Toluene steam reforming C 7 H 8 +7 H 2 O→7 CO+11 H 2 (1) Toluene steam dealkylation C 7 H 8 +H 2 O→C 6 H 6 +CO+2 H 2 (2) Water gas shift (WGS): CO+H 2 O↔CO 2 +H 2 (3) Thermal cracking C 7 H 8 →x C n H m +z H 2 (4) C 6 H 6 →x C n H m +z H 2 (5) Hydrodealkylation C 7 H 8 +H 2 →C 6 H 6 +CH 4 (6) Toluene dry reforming C 7 H 8 +7 CO 2 →14 CO+4 H 2 (7) Boudouard reaction: C+CO 2 ↔2 CO (8) Methanation CO+3 H 2 ↔CH 4 +H 2 O (9) Coke heterogeneous gasification: C+H 2 O→CO+H 2 (10) Amongst these reactions, the most important ones are steam reforming (Eq. (1)) and water gas shift (WGS) (Eq. (3)). The selection of temperature and catalyst determines the extent of these reactions and the selectivity towards the different products [7,15–18]. Tar catalytic conversion methods are classified as in-situ (or primary) and post-gasification (secondary) ones [14,19]. In the former, the tar reduction occurs during the gasification stage, with the catalyst being located in the gasifier itself. In the secondary approach, the gas produced in the gasifier is treated downstream in a secondary reactor where the catalyst is placed. Regardless the strategy followed, essential aspects conditioning the gasification process are those involving the reactor configuration, operating conditions and type of catalyst [20]. Fluidized bed reactors are one of the most developed technologies for biomass gasification, which require an appropriate catalyst in terms of activity and stability in order to reduce the tar content to 2 g m −3 and avoid the need of a more expensive secondary catalytic reactor downstream [21,22]. A wide range of materials with significant activity for cracking and reforming of heavy aromatic compounds have been investigated as primary catalysts [12,14]. Natural minerals, such as olivine and calcined dolomite, have been widely used in the steam gasification in fluidized beds, as they are active for tar removal, apart from being inexpensive and abundant [23]. Acid catalysts, such as alumina or zeolites, have also been used (prior to and after metal impregnation) as catalysts for tar abatement [24]. Nevertheless, the performance of all these primary catalysts can be greatly improved by metal phase addition [19,25–28]. Thus, support features, such as mechanical (resistance to attrition), physico–chemical (surface area, porosity, acidity, composition and density) and catalytic ones (activity / selectivity and stability) play a relevant role in the metal-support interactions, as well as in the reforming reaction mechanism itself [29]. From a catalytic point of view, nickel is known to be the most interesting metal phase for reforming applications [30]. Ni-based catalysts have been widely applied in the steam reforming of biomass tars due to both their high activity for breaking C – C and O – H bonds and performance in terms of H 2 production [7,12,13,31–35]. However, their main drawbacks are related to their rapid deactivation, mechanical fragility and high cost compared to natural minerals or alumina [36]. Currently, use of iron as an active phase is gaining more attention for tar reduction due to its lower cost, abundance and lower toxicity compared to nickel [37]. Iron is known to be an active species for aromatic hydrocarbon destruction (breakage of C – C and C – H bonds), as well as for the WGS reaction. In fact, it has been proven effective for the aforementioned reactions in different oxidation states [24,38–40]. Therefore, iron impregnation of olivine, dolomite or Al 2 O 3 seems to be interesting for synthetizing in-bed primary catalysts for gas–solid contact reactors, such as fluidized or spouted beds, from both economic and environmental perspectives. Nevertheless, the lower activity of the iron species with respect to the Ni ones requires higher amounts of dopant, generally in the 10–30 wt% range [41]. Accordingly, the aim of this work is to analyse the performance of olivine, dolomite and γ –Al 2 O 3 as primary catalysts, as well as the effect the impregnation of each catalyst with 10 wt% Fe has on the elimination of toluene, which has been selected as the model compound of biomass gasification tar. Furthermore, a detailed characterization of the fresh and deactivated Fe-doped catalysts has been carried out in order to determine the main deactivation mechanisms in this process. The results obtained will provide essential information for the selection of optimal primary catalysts for biomass gasification in the bench-scale unit equipped with an improved spouted bed reactor developed by our research group [21,42,43]. Furthermore, the results obtained may also be extrapolated to industrial gasification reactors, which are mainly fluidized beds. This study addresses multiple aspects that have not been jointly approached in the literature, as are catalysts preparation and characterization, influence of temperature, catalyst performance at zero time on stream, stability of Fe-loaded catalysts and the main deactivation causes. 2. Experimental 2.1. Catalyst preparation and characterization Six catalysts have been tested in the toluene steam reforming process. Three of them (olivine, dolomite and γ –alumina) are primary catalysts, whereas the other three are those obtained by impregnating the aforementioned primary catalysts with Fe, i.e., Fe/olivine, Fe/ dolomite and Fe/Al 2 O 3 . Besides, runs with silica sand were carried out for comparison purposes. Minerals Sibelco supplied the olivine and dolomite, and Alfa Aesar the γ-Al 2 O 3 . These three primary catalysts provided satisfactory results in a previous study of biomass gasification in a fountain confined conical spouted bed reactor (CSBR), as they allowed reducing tar formation, as well improving the yield and composition of the syngas [21,42,43]. The catalyst particles were sieved in order to retain the fractions within the ranges of 90–150 μ m for olivine, 150–250 μ m for dolomite and 250–400 μ m for γ - Al 2 O 3 , which allow attaining similar fluidization regimes with these materials of different densities. Prior to use, dolomite was calcined at 900 ◦C for 4 h in a muffle oven in order to complete the decarboxylation of calcium and magnesium carbonates. The Fe loaded catalysts were prepared by wet impregnation of the supports with an aqueous solution of Fe(NO 3 ) 3 ⋅9H 2 O (Panreac AppliChem, 98 %). The amount of saline precursor added was that corresponding to the desired final catalyst composition. The concentration of Fe was fixed at 10 wt% in order to compare the catalytic activity and selectivity of the three catalysts for same amount of metal loaded. Subsequent to the impregnation process, the prepared catalysts were dried at 100 ◦C for 24 h and then calcined at 1000 ◦C for 4 h. Given that the catalytic activity of iron species generally increases with their reduction state (Fe 2 O 3 <Fe 3 O 4 <FeO <Fe(0)) [26], these ironimpregnated catalysts were used once they had been subjected to an M. Cortazar et al. Fuel 317 (2022) 123457 3 ex situ reduction process at 850 ◦C for 4 h under 10 vol% H 2 stream, which ensured full reduction of ferric oxides into their metallic phase. The particle sizes of the Fe loaded catalysts were the same as those of their primary counterparts. The physical properties of the catalysts were determined by N 2 adsorption–desorption in a Micromeritics ASAP 2010 instrument. Based on the information of these isotherms, the catalysts features, such as those involving specific surface area and porous structure (average pore size and pore volume), were calculated by the Brunauer–Emmett–Teller (BET) method. Prior to the analysis, and in order to remove any impurity, the samples were degassed at 150 ◦C until a pressure below 2⋅10 −3 mmHg was reached. The chemical composition (wt%) of each catalyst was measured by X-ray fluorescence (XRF). More detailed information about the XRF methodology can be found elsewhere [43]. The temperature-programmed reduction (TPR) of the catalysts was carried out in an AutoChem II 2920 Micromeritics, which allowed determining the catalyst reduction temperature before using it. This method consists in exposing the solid to a reducing gas flow of 10 vol% H 2 /Ar, while temperature is increased with a constant heating rate of 5 ◦C min −1 from ambient one to 900 ◦C. The reduction temperature of each catalyst was ascertained by monitoring the H 2 consumed. The crystalline structure of the fresh and deactivated catalysts was analyzed using X-ray powder diffraction (XRD) patterns. A Bruker D8 Advance diffractometer with Cu K α 1 radiation was used to conduct XRD. The detailed procedure followed is described elsewhere [44]. The metal crystallite size was calculated by using the Scherrer formula. Metal dispersion was calculated from metal crystallite size using the equation D (%) =97.1/d (nm) and assuming that the size of Fe atom is the same as that of Ni atom, as reported elsewhere [45,46]. The values of total acidity of the catalysts have been obtained by monitoring the differential adsorption of NH 3 at 150 ◦C using simultaneously calorimetry and thermogravimetry in a Setaram TG-DSC 111 equipment. The amount of coke deposited on the used catalysts was determined by temperature-programmed oxidation (TPO) in a thermobalance (TGA Q5000TA Thermo Scientific). This TGA is connected on-line to a Blazer Instruments Mass Spectrometer (Thermostar) and the procedure followed to determine the coke deposited on each sample is as follows: (i) signal stabilization with He stream at 100 ◦C for 30 min, and (ii) a ramp of 5 ◦C min −1 to 800 ◦C in a stream of O 2 diluted in He, with this temperature remaining constant for 30 min in order to ensure full coke combustion. 2.2. Experimental equipment and procedure The experiments of toluene conversion with the different catalysts were performed in an Inconel fluidized bed reactor (300 mm in length and 10 mm in internal diameter), as shown in Fig. 1. The reactor is located within a radiant oven, which provides the heat for operating up to 900 ◦C. The temperature was measured and recorded by means of two K-type thermocouples, with one being located inside the reactor, approximately in the middle zone of the bed, and the other one close to the wall of the electric oven. The water for generating the steam and toluene were introduced by means of a high-pressure pump (ASI 521) and a syringe pump (PHD 4400), respectively. Their pumping flowrates were maintained constant in all the runs, with the values being 0.24 mL min −1 for water and 0.06 mL min −1 for toluene, which correspond to a steam/toluene ratio (S/T) of 4 and a molar steam/carbon (S/C) ratio of 3.35. Prior to feeding into the reactor, these two compounds were pumped separately into an evaporation system at 350 ◦C, which ensures their full vaporization. This plant is also provided with a nitrogen mass flow meter (Brooks SLA5800) that allows feeding up to 1 L min −1 . In fact, a nitrogen flow rate of 300 mL min −1 was used as fluidizing agent during the heating process prior to the reaction. The gaseous stream leaving the reactor was passed through a heater, whose temperature was kept at 300 ◦C in order to prevent the condensation of the products before entering the on-line analysis system. Then, the volatile stream circulated through a condensation device consisting of two coalescence filters, which ensured total condensation and retention of the non-reacted steam and toluene, as well as toluene derived products. This study deals with the effect of operating conditions in a catalytic process for tar elimination process, i.e., reforming temperature (in the 800–900 ◦C range), catalysts type (olivine, dolomite and alumina, as well as their counterparts with Fe impregnation) and catalyst stability. Olivine was chosen to analyse the effect of temperature, whereas 850 ◦C was established as the suitable operating temperature to study the influence of catalyst type and time on stream. The effect of reaction time was studied for the Fe loaded catalysts in the 5–115 min range in order to assess the evolution of catalyst activity and stability. Given that the density of the primary catalysts differs significantly (3300 kg m −3 for olivine, 1666 kg m −3 for alumina and 1275 kg m −3 for dolomite), and in order to operate under the same hydrodynamic conditions in the fluidized bed reactor, the same bed volume was used in all experiments. Accordingly, as mentioned above, suitable particle sizes were chosen. Thus, 3.8 cm 3 of the corresponding catalyst (or sand in case the experiment was carried out without catalyst) were placed in the bed in all cases, corresponding to a gas hourly space velocity (GHSV) of 820 h −1 . Experiments at zero time on stream were repeated at least 3 times to ensure reproducibility of the results and the carbon mass balance closure was above 95 % in all runs. 2.3. Product analysis The analysis of the volatile stream leaving the reactor was conducted on-line by means of a GC (Agilent 7890) provided with a flame ionization detector (FID). The sample was injected into the GC prior to condensation by means of a line maintained at 280 ◦C in order to avoid the condensation of heavy tar compounds. The analysis of the noncondensable gases (after separating the tars from the gaseous stream in the condensation system) was carried out by means of a micro GC Fig. 1. Schematic diagram of the toluene steam reforming laboratory unit. M. Cortazar et al. Fuel 317 (2022) 123457 4 (Agilent 4900). The three independent modules with different columns (molecular sieve, porapak and plot alumina) allowed identifying and quantifying the gaseous products previously calibrated. This analysis methodology allowed a detailed quantification of the entire product stream. 2.4. Reaction indices The conversion and product yields were taken as reaction indices to monitor and assess process performance. The carbon conversion of toluene was defined as the moles of carbon in the gaseous product stream divided by the moles of carbon in the toluene feed (Eq. (11)). Note that the moles of CO, CO 2 and C 1 -C 4 hydrocarbons formed (corresponding to the total amount of carbon moles in the gas) have been determined from the micro-GC analysis, whereas the moles of carbon in the feed were calculated based on the total amount of toluene introduced into the reactor (total volume of toluene injected in the run). Cconversion(%) = moles of carbon in the product gas moles of carbon in the feed ⋅100 (11) The product yields were calculated as the ratio between the grams of each product (H 2 , CO, CO 2 and CH 4 ) in the gaseous stream and the grams of the model compound in the feed: Yield(wt%) = g of the compound in the product gas g of model compound in the feed ⋅100 (12) Moreover, H 2 potential was also determined as the ratio between the concentration of H 2 in the effluent gas and the maximum allowed by stoichiometry: H2potential(%) = moles of H2in the product gas maximum moles of H2allowed by stoichiometry (13) The maximum number of H 2 moles allowed by stoichiometry was calculated by considering toluene reforming reaction and that of WGS. Thus, H 2 potential is defined based on the maximum number of H 2 moles obtained when toluene is fully reformed to CO 2 and H 2 . 3. Results and discussion 3.1. Fresh catalyst characterization Table 1 shows the physical properties (specific surface area, pore volume and average pore diameter) and chemical composition of the primary catalysts and those impregnated with Fe. As observed, olivine has the lowest specific surface area (1.92 m 2 g −1 ) and pore volume (0.002 cm 3 g −1 ) due to its non-porous structure. After impregnation with Fe(NO 3 ) 3 ⋅9H 2 O solution, the specific surface area and the pore volume of dolomite and Al 2 O 3 decreased mainly due to metal deposition, as it blocks some of the micropores of the catalysts. According to Kumar et al. [47], the presence of iron on alumina accelerates the shrinkage of alumina and transforms the alumina from gamma into other phases, which decreases the surface area because Fe 2 O 3 particles act as heterogeneous nucleation sites for α -Al 2 O 3 particles at high temperature. Nevertheless, the opposite trend was observed in the Fe/ olivine, i.e., the specific surface area increased due to the deposition of Fe on the external surface area, and the pore volume and average pore size became larger, which suggests the collapse of the inter-pore structure of olivine. Note that the same trend has been observed for metal impregnation on supports with low porosity surfaces [22,48,49]. Apart from the impregnation process, the high calcination temperature also contributes to reducing the BET surface area and porosity of the Fe/ Al 2 O 3 catalyst, although to a lesser extent. In a previous study [50], the same Al 2 O 3 used in this study was calcined with air at 1000 ◦C during 5 h and its BET surface area and pore volume reduced to 87 m 2 g −1 and 0.38 cm 3 g −1 , respectively. Dolomite is a calcium magnesium carbonate, i.e., CaMg(CO 3 ) 2 , and therefore carbonates are decomposed into CaO and MgO in the calcination, which are the main constituents in the calcined dolomite, as shown in Table 1. Moreover, the XRF revealed that there is a high content of Fe in the Fe/olivine. In fact, the content of Fe in the raw olivine was of around 5.3 wt% and after impregnation, the Fe amount in the catalyst increased significantly to 17 wt%, which confirmed that the metal content was close to that corresponding to the impregnation (~10 wt%) plus that in the original olivine. In the other two catalysts, namely Fe/Al 2 O 3 and Fe/dolomite, the initial Fe content was negligible and after the impregnation increased up to 9.9 and 9.3 wt%, respectively. Thus, the Fe content of the three studied catalysts is consistent with the targeted metal loading of 10 wt%. Table 2 shows the metal dispersion of each catalyst which was estimated based on the metal crystallite size obtained by XRD analysis (by applying Debye-Scherrer equation). As observed, the highest metal dispersion is attained for Fe/Al 2 O 3 (2.6 %), whereas the poorest value is for dolomite (0.5 %). This result confirms that the physical structure of the support plays an essential role in the dispersion of the metal phase; that is, the support with the highest surface area as that of Al 2 O 3 leads to the highest metal dispersion. Table 1 Properties of primary and Fe impregnated catalysts. Olivine Fe/Olivine γ-Al 2 O 3 Fe/Al 2 O 3 Calcined Dolomite Fe /Dolomite Physical properties S BET (m 2 g −1 ) 1.92 3.75 100.00 12.48 17.42 3.55 V pore (cm 3 g −1 ) 0.002 0.017 0.42 0.059 0.05 0.009 d pore (Å) 78 234 167 206 113 162 Chemical properties MgO (wt%) 48.79 36.98 – 0.23 43.61 32.15 SiO 2 (wt%) 43.18 37.20 0.02 – 0.12 0.11 Fe 2 O 3 (wt%) 7.68 24.39 – 14.13 0.02 13.21 CaO (wt%) 0.12 0.17 – 0.14 56.07 48.53 Al 2 O 3 (wt%) 0.04 0.43 99.98 81.02 0.15 0.26 Na 2 O (wt%) 0.06 0.06 – – 0.01 0.03 TiO 2 (wt%) 0.02 0.03 – 0.12 0.02 0.03 MnO (wt%) 0.11 0.10 – – – – Acidity Total acidity (µmol NH 3 g cat−1 ) 2.4 8.8 80.0 11.4 8.7 10.5 Table 2 Metal dispersion (%) calculated from metal crystallite for the three Fe impregnated catalysts. d Fe XRD a (nm) Fe dispersion (%) Fe/Al 2 O 3 38 2.6 Fe/olivine 68 1.4 Fe/dolomite 193 0.5 a Calculated from the full width at half height of the Fe 0 (110) diffraction peak at 2θ =44◦in the XRD profiles using the Scherrer equation. M. Cortazar et al. Fuel 317 (2022) 123457 5 The XRD patterns of the primary catalysts and Fe reduced catalysts are shown in Fig. 2a and 2b, respectively. As observed, the three Fe doped catalysts show an intense peak of the metal iron phase at 2θ =44◦ and two smaller ones at 2θ =65◦and 82◦. Note that iron oxide phases were not detected in these catalysts, which is evidence of their full reduction. In both olivine (Fig. 2a) and Fe/olivine (Fig. 2b), the main crystalline phases observed are those corresponding to olivine (Mg 1.81 Fe 0.19 ⋅(SiO 4 )) and enstatite (MgSiO 3 ). Further diffractogram of unreduced Fe/olivine catalyst can be found elsewhere [22]. Regarding Fe/dolomite (Fig. 2b), apart from the metal iron phase, those of Ca (OH) 2 , CaO and MgO were also observed, with all of them being derived from the calcination of calcium magnesium carbonate, which is the main mineral species in the dolomite [51]. These last three phases (Ca(OH) 2 , CaO and MgO) were also observed in the XRD diffractogram of calcined dolomite (Fig. 2a). These alkaline earth oxides (CaO and MgO) containing Lewis basic sites may promote adsorption and migration of H 2 O and OH groups on the catalyst surface, and therefore promote carbon gasification and reduce carbon deposition [52]. The Ca(OH) 2 diffraction peaks are evidence that CaO (a highly hygroscopic compound) absorbed humidity from the ambient and formed Ca(OH) 2 . In the Fe/Al 2 O 3 catalyst, typical diffraction peaks corresponding to the Al 2 O 3 support were detected, as well as hercynite (FeAl 2 O 4 ), whose diffraction lines are located at 2θ =31◦, 36◦, 51◦, 59◦and 64◦. The high calcination temperature used (1000 ◦C) allowed the formation of hercynite spinel (FeAl 2 O 4 ), which occurs at temperatures above 600 ◦C by the interaction between Fe species (Fe 0 , FeO and Fe 3 O 4 ) and Al 2 O 3 , following the reaction mechanism reported in the literature [53,54]. Moreover, comparing the XRD patterns of Al 2 O 3 before and after impregnation and calcination stages, there is a phase change from γ-Al 2 O 3 to a more stable one, which is probably the most stable one ( α -phase) due to the high temperature of calcination used (1000 ◦C). The peaks assigned to Al 2 O 3 in the Fe loaded catalyst in Fig. 2b are clear and sharp, which is evidence of its high crystallization degree, whereas the peaks assigned to Al 2 O 3 in Fig. 2a are broad and low, thereby suggesting an amorphous structure with a small crystallization degree. Note that the same diffraction peaks than those observed for Al 2 O 3 crystalline phases in Fig. 2a and b have been reported in the literature and correspond to γ-Al 2 O 3 and α -Al 2 O 3 , respectively [55,56]. Therefore, phase transformation is the consequence of the thermal degradation of the support, which affects adversely the physical properties of the catalyst by reducing catalyst surface area, thereby reducing catalyst activity. Several authors have called this process support sintering [35,57]. The temperature programmed reduction (TPR) profiles of calcined Fe/olivine, Fe/dolomite and Fe/Al 2 O 3 catalysts are shown in Fig. 3. Given that metal iron is expected to be the active phase for breaking C – C and C – H bonds [24,58], the reducibility of the catalysts is of great Fig. 2. XRD patterns of primary catalyst (a) and Fe impregnated ones (b). Crystalline phases: (+) (Mg 1.81 Fe 0.19 (SiO 4 )); (o) Enstatite (MgSiO 3 ); (∇) Fe 0 ; (◆) Al 2 O 3 ; (□) MgO; (•) CaO; (❖) Hercynite (FeAl 2 O 4 ); (✦) Calcium hidroxide (CaOH 2 ). M. Cortazar et al. Fuel 317 (2022) 123457 6 relevance. According to the literature [26,59] the reduction of Fe 2 O 3 generally proceeds in two steps, as are: the reduction of Fe 2 O 3 to Fe 3 O 4 in the 350–500 ◦C range and the reduction of Fe 3 O 4 to metal Fe in the 500–900 ◦C range. However, according to certain studies, the intermediate FeO is formed in the reduction from Fe 3 O 4 to Fe 0 [60,61]. These two regions associated with two or three reduction steps from Fe 2 O 3 are observed in the three reduced catalysts, although differences in the interactions between the iron and the supports shifted the location of the peaks. In the TPR profile of the Fe/olivine, a broad reduction zone between 350 and 700 ◦C is observed with 3 peaks. The first two (at 470 and 530 ◦C) are associated with the reduction of Fe 2 O 3 and Fe 3 O 4 /FeO, respectively, whereas the latter peak above 600 ◦C is due to the Fe atoms that migrated into the olivine support to form a very stable MgFe 2 O 4 spinel phase [62]. Peaks at 380 ◦C and 500 ◦C appear in the Fe/dolomite, which are characteristic of iron species reduction, but there is also a broad peak at 750 ◦C, which corresponds to the reduction of Fe 3+ from the calcium iron oxide (srebrodolskite, Ca 2 Fe 2 O 5 ) to Fe, as was suggested by Zamboni et al. [63,64]. These authors observed the formation of this phase when iron nitrate was used in the wet impregnation of dolomite. In this study, no evidences of Ca 2 Fe 2 O 5 are observed in the XRD diffractogram (Fig. 2), probably due to its low crystallinity . In the Fe/Al 2 O 3 catalyst, apart from the two peaks identified at 380 and 580 ◦C, which are associated with the reduction of iron species (Fe 2 O 3 , Fe 3 O 4 and FeO) , a third broad reduction zone appears between 700 and 900 ◦C, which is attributed to the reduction of iron aluminates (FeAl 2 O 4 ), also identified in the XRD spectra [65]. Different authors suggested that the presence of alumina stabilizes Fe 2 O 3 phase and the reduction goes through the formation of FeAl 2 O 4 spinel, whose reduction occurs above 700 ◦C [66,67]. 3.2. Role of temperature in the tar conversion on olivine catalyst The influence of temperature on toluene abatement on olivine catalysts is displayed in Fig. 4. Fig. 4a shows the evolution of carbon conversion and H 2 potential. As observed, temperature has a great influence on carbon conversion and H 2 potential, since their values increase from 3.6 and 2.6 % at 800 ◦C to 46.0 and 23.6 % at 900 ◦C, respectively. This increase in both parameters is attributed to the endothermic nature of the toluene reforming reactions, as well as of those involving decomposition and dehydrogenation, as all of them are promoted at high temperatures [68]. The same trend of carbon conversion and H 2 potential with temperature on olivine catalysts was observed by other authors in the tar steam reforming [58,69]. The yields of the compounds in the product stream is displayed in Fig. 4b. An increase in temperature leads to higher yields in gaseous compounds (including benzene) due to the promotion of both reforming and cracking reactions, with the highest yields being those of CO and CO 2 at 900 ◦C (49.8 and 26.8 wt%, respectively). The yield of CH 4 increases with temperature, but it is lower than 2.2 wt% at the three temperatures studied. It should be noted that the yield of C 2 -C 4 hydrocarbons is hardly noticeable (below 0.01 wt%), and has not therefore been included in Fig. 4b. CH 4 is mainly formed from dealkylation of the methyl group in the toluene structure and, to a minor extent, from the methanation of CO [8]. However, steam reforming of CH 4 prevails over these reactions, since the content of CH 4 in the products is very low [12]. The presence of an undesired compound (benzene) is due to incomplete decomposition of toluene [70], which is confirmed in Fig. 4b, where benzene yield increases from 0.6 wt% at 800 ◦C to 12.6 wt% at 900 ◦C at the expense of a decrease in toluene yield. Several reactions, such as steam dealkylation (Eq. 2), thermal cracking (Eqs. 4–5) or hydrodealkylation of toluene (Eq. 6) lead to the formation of benzene (all of them enhanced at high temperatures) [11,15,17,71]. However, the small amount of CH 4 in the product stream is evidence that hydrodealkylation reaction (Eq. 6) is not significant [72]. It should be noted that the benzene produced from the aforementioned reactions can undergo reforming reactions to produce further CO and H 2, although these reactions are limited due to benzene stability [11,12]. The yield of polycyclic aromatic hydrocarbons (PAHs, referred to the compounds heavier than toluene) also increases with temperature due to the promotion of condensation reactions of lighter tars. However, the low yield of these PAHs (below 1.2 wt% in the whole range of temperatures studied) is evidence that the extent of these reactions is almost negligible, probably due to the presence of steam [21,73]. Swierczynski et al. [3] also observed a yield of around 6 wt% of benzene and 14 wt% of polyaromatics in the product stream of toluene steam reforming at 850 ◦C when they used olivine as primary catalyst. Fig. 4c displays the gas composition in the 800–900 ◦C range. It can be observed that the effect of temperature on the gas composition is not very pronounce above 850 ◦C, i.e., the concentration hardly changes above this temperature. Between 800 and 850 ◦C, certain trends are observed when temperature is increased, as are: a slight decrease in H 2 and CO 2 concentrations (from 69.1 to 66.2 vol% and from 8.1 to 6.8 vol %, respectively) and an increase in that of CO (from 21.8 to 25.5 vol%). This result is explained by the promotion of the reverse WGS reaction due to its exothermic nature. The same trend with temperature was observed in other studies of catalytic reforming of tar model compounds, with this effect being attributing to the exothermic nature of the WGS reaction [68,74]. 3.3. Comparison of primary catalysts performance In order to study the performance of primary catalysts, toluene conversion on olivine, dolomite and alumina was monitored at 850 ◦C and the results obtained are displayed in Fig. 5. The effect of thermal cracking was ascertained by comparing the results of carbon conversion (Fig. 5a), product yields (Fig. 5b) and concentration of gaseous Fig. 3. TPR profiles of Fe/Al 2 O 3 , Fe/dolomite and Fe/olivine catalysts. M. Cortazar et al. Fuel 317 (2022) 123457 7 compounds (Fig. 5c) obtained with the catalysts and those obtained with inert sand. As observed, the presence of any catalyst improves the overall efficiency of the process by increasing carbon conversion and the yields of gaseous compounds, especially those of H 2 , CO and CO 2 , as well as reducing that of toluene. This improvement over the results obtained with inert sand is associated with the promotion of steam reforming (Eq. 1), cracking (Eqs. 4–5) and WGS reactions (Eq. 3). The presence of primary catalysts also promotes steam dealkylation (Eq. 2) and thermal cracking (Eq. 4) reactions, since the concentration of benzene in the product stream is higher than that obtained with sand. Comparing the efficiency of the primary catalysts (Fig. 5a), Al 2 O 3 leads to the highest conversion (58.4 %) followed by dolomite (39.1 %). However, the H 2 potential with both catalysts is similar (28.5 % for Al 2 O 3 and 28.9 % for dolomite). This latter result can be explained by the lower activity of Al 2 O 3 and the higher of dolomite in the WGS. Thus, the higher activity of dolomite in the WGS reaction is related to CaO and MgO basic sites, with activity being higher as the Ca/Mg ratio is increased [75,76]. Furthermore, the presence of CaO and MgO also explains the higher yield of benzene at the expense of lowering that of toluene [77,78]. Moreover, olivine has the smallest influence on the toluene steam reforming, since it provided the lowest carbon conversion and H 2 potential values. In this case, although the presence of Fe promotes reforming reactions, the low BET surface area (1.91 m 2 g −1 ) and pore volume (0.002 g cm −3 ) are the factors leading to the low efficiency of this catalyst in the toluene elimination process. Studies reported in the literature confirm that dolomite and Al 2 O 3 were more active than olivine for reducing the amount of tar derived from biomass gasification, as the extent of the WGS reaction is enhanced with dolomite [43,79]. 3.4. Effect of Fe incorporation into the primary catalysts Fig. 6 compares the parameters involving toluene conversion (carbon conversion and H 2 potential (a), product yields in the outlet stream (b) and the concentration of gaseous compounds (c)) for the Fe loaded catalysts. Fig. 6a reveals that Fe incorporation into the primary catalysts leads to higher carbon conversion and H 2 potential than those on the primary catalysts in all cases, Fig. 5a, which is evidence of their higher catalytic activity for toluene reforming. Thus, on the one hand, it is well stablished that metal iron is active for C – C and C – H bond breakdown, which enhances hydrocarbon reforming and cracking reactions [58,80]. On the other, the addition of Fe promotes the WGS reaction because the adsorption of water molecules on the catalyst active sites is favoured, thus leading to higher H 2 yields [81]. This improvement is especially remarkable with olivine, whose carbon conversion and H 2 potential increases from 18 and 10.5 % to 73 and 31.9 %, respectively. As occurred with primary catalysts, that of Fe/Al 2 O 3 provided the best results in terms of carbon conversion (87.6 %) and H 2 potential (38 %) (Fig. 6a). However, the trends were reversed for Fe/olivine and Fe/ dolomite after Fe incorporation, attaining higher carbon conversion in the former. This result is closely related to the change in the surface area of the catalysts caused by the impregnation, which definitely affects metal dispersion. As observed in Table 1, the surface area increased in the olivine when Fe was introduced, whereas it significantly decreased in the dolomite (from 17.42 to 3.55 m 2 g −1 ). Furthermore, the results in Fig. 4. Effect of temperature on carbon conversion and H 2 potential (a), product yields and unreacted toluene fraction in the outlet stream (b), and concentration of the gaseous stream (c). M. Cortazar et al. Fuel 317 (2022) 123457 8 Table 2 confirm the better dispersion of Fe on the olivine than on the Fe/ dolomite, which suggests that the active sites are more accessible for the reactants in the former, as all the iron is located on the catalyst surface. This implies a higher catalytic activity of Fe/olivine, which explains the better results of carbon conversion on this catalyst than on Fe/dolomite, whose metal dispersion is the poorest. Comparing the product yields shown in Fig. 6b, the highest yields of CO and CO 2 (mainly derived from the reforming and WGS reactions, respectively) and benzene (a cracking product) are obtained on the Fe/ Al 2 O 3 catalyst, whereas that of toluene is the lowest (below half of those obtained on Fe/olivine or Fe/dolomite). Note that Fe acts as the active phase for the reforming and WGS reactions, whereas the alumina support provides the acidity required for cracking reactions, i.e., the combination of both provides Fe/Al 2 O 3 catalyst with the highest activity for these reactions. Moreover, a comparison of Fig. 6b with Fig. 5b shows that the yield of benzene increases greatly when Fe is added to the primary catalysts. It seems that the presence of Fe mainly catalyzed the conversion of toluene to benzene. Some studies suggested that temperatures higher than 800 ◦C increase the hydrodealkylation activity for the steam reforming of toluene on iron-based materials [72,82], whereas other researches concluded that the activity of iron-based materials leads to the decomposition of large tar compounds into small fragments of carbon species, which subsequently form benzene [83]. Therefore, it can be concluded that the higher benzene content is a combined effect of cracking and hydrodealkylation of toluene molecules on Fe active sites. The higher CH 4 yields observed on Fe loaded catalyst than on primary catalysts also confirms this hypothesis. Fig. 6c displays the gas composition obtained with the three Feimpregnated catalysts. A comparison of these results with those for primary catalysts (Fig. 5c) shows the relevance of metal iron in the WGS reaction (Eq. 3), since the concentration of CO 2 greatly increased in all the cases, whereas that of CO reduced. This is consistent with previous studies in the literature, in which a high activity of Fe is reported in the WGS reaction [38,40,84]. Analysing Fe loaded catalysts, Fe/Al 2 O 3 led to the lowest concentration of CH 4 and CO 2 and the highest of H 2 and CO, which is evidence of a high extent of steam and dry reforming of hydrocarbons (Eq. 1 and 7). According to Adnan et al. [85,86], this fact is attributed to the basic sites of Fe/Al 2 O 3 catalysts, which promote endothermic CO 2 reforming of hydrocarbons. The differences observed among these Fe-impregnated catalysts are the consequence of various factors. As previously stated, one of the most influential factor is related to the metal dispersion on the catalyst support, which plays a key role in the initial catalyst activity. A suitable metal-support interaction enhances the migration of metal crystallites, thereby obtaining a better dispersion of Fe on the support [26]. Furthermore, the physical structure of the support greatly influences the dispersion of the metal phase, as shown in Table 2, in which the highest Fe dispersion was obtained for Al 2 O 3 (the support with the highest BET surface area and pore volume). The results in Fig. 6 confirm that the better surface properties of the Al 2 O 3 support promote the dispersion of Fig. 5. Effect of primary catalysts on carbon conversion and H 2 potential (a), product yields and unreacted toluene fraction in the outlet stream (b), and concentration of gaseous compounds (c). M. Cortazar et al. Fuel 317 (2022) 123457 9 the active phase, and therefore lead to higher catalyst activity. Besides, the higher dispersion of Fe on olivine also explains the higher carbon conversion than on Fe/dolomite. Another factor is related to the activity of the support for cracking and/or reforming reactions, which is directly linked to its acidity [30,57]. Thus, the porous structure of olivine and dolomite barely have micro or mesopores, whereas alumina has a more developed porous structure, as shown in Table 1. Adnan et al. [85] suggested that toluene conversion reactivity is dominated by strong acid sites in the catalyst, which are directly attached to the surface of the catalyst. Thus, a higher surface area of the catalyst increases the number of strong sites available to contact with toluene, thereby leading to a higher acidity of the catalysts, and consequently to a higher conversion of toluene, as is the case of Fe/Al 2 O 3 , which has the highest acidity (Table 1) of the three Fe loaded catalysts [87]. Besides, Adnan et al. [88] stated that a higher content of Fe in the catalyst also promotes catalyst acidity, and therefore toluene conversion. Comparing the acidity of primary and Fe doped catalysts (Table 1), the presence of Fe increases the acidity of Fe/olivine and Fe/dolomite catalysts from 2.4 to 8.8 and from 8.7 to 10.5 µmol NH 3 g cat−1 , respectively, which explains the higher toluene cracking capability of Fe doped ones. Regarding the acidity value of Fe/Al 2 O 3 (11.4 µmol NH 3 g cat −1 ), it is much lower than that of the raw γ-Al 2 O 3 . Indeed, as previously stated, the reduction in BET surface area caused by the calcination and impregnation stages leads to the blockage of some pores and reduces the number of acid sites available, thus reducing the total acidity of the catalyst. However, comparing Fig. 5b and 6b, benzene yield is higher when Fe/Al 2 O 3 is used than when the primary Al 2 O 3 is used, which suggests that the cracking activity of Fe/Al 2 O 3 is higher. This result is explained by the combination of two issues. On the one hand, as was previously stated, the better performance of Fe for reforming and WGS reactions leads to higher H 2 partial pressures in the reaction environment, thus promoting hydrodealkylation reactions (Eq. 6) which lead to higher benzene contents. On the other hand, the real acidity of γ-Al 2 O 3 under reaction conditions is much lower than that given in Table 1, as the high temperatures used in this study (850 ◦C) and the presence of steam accelerate the collapse of the porous structure and the transformation of γ-alumina into other more stable phases, as stated elsewhere [47]. Thus, the blockage of pores and the transformation of γ-phase into other ones (δ, θ or α ) reduces the number of acid sites available, and therefore its cracking activity. Other important issue involving catalytic activity is the reduction state of the iron species, with activity being higher as Fe species are further reduced (metal Fe is the most active phase). Thus, the XRD patterns in the three fresh catalysts reveal the presence of metal Fe, whereas the presence of other species with different reduction states, such as Fe 2 O 3 , Fe 3 O 4 or FeO, was not initially observed (Fig. 2b). This is an evidence that the difference in the catalytic activities of Fe impregnated catalysts is mostly attributed to the interactions between the metal iron and the supports, as well as their physical structure. Thus, the better properties of Al 2 O 3 (it acts as a textural promoter preventing the fast sintering of the iron metal, as well as stabilizing active sites on its surface) lead to better dispersion of the Fe oxide phase, and therefore better Fig. 6. Effect of Fe impregnated catalysts on carbon conversion and H 2 potential (a), product yields and unreacted toluene fraction in the outlet stream (b), and concentration of gaseous compounds (c). M. Cortazar et al.